Using machine learning to analyze R-loop imaging data

Berkant Cünnük, MSc student – University of Manitoba

Supervisor – Margherita Maria Ferrari, Department of Mathematics, University of Manitoba
Collaborator – Frédéric Chédin, Department of Molecular and Cellular Biology, University of California, Davis

Project Description: R-loops are three-stranded complexes made of a DNA:RNA hybrid and a single strand of DNA. These complexes may act as cellular regulators or as source of genomic instability, which underlies the increasing interest on R-loops from both experimentalists and mathematicians. Previous work by the Chédin Lab showed that the arrangement of the single-stranded DNA in an R-loop dictates the overall R-loop configuration; in fact, three distinct configurations were (manually) identified by analyzing atomic force microscopy (AFM) imaging data [Carrasco-Salas et al. Nucleic Acids Res. 47(13), 6783-6795 (2019)]. The goal of this project is to develop a computational pipeline to classify and distinguish R-loop configurations by analyzing R-loop AFM imaging data with machine learning techniques. The results will provide insights for modeling the entanglement of the strands in an R-loop, an aspect that is currently not well understood.